How I Use AI, and Why I'm Publishing Here

I use AI tools when I write, and I say so. Every essay on this site carries a short process tag under the title and a fuller note at the end, so you can see how it was made: what I wrote, what Claude wrote, and who made the decisions.

As a writing teacher, I don’t think the interesting question is whether a sentence passed through a language model. It’s whether someone did the thinking: drafting, rereading, judging, revising, and doing it again. That iterative work can happen with AI in the room. It’s the part I want to protect in my students’ writing and in my own, and I’d rather show it than assert it.

How each essay is made

I work in a few different ways, depending on the piece. Each essay is labeled with one of these.

Directed drafting

I talk through my ideas out loud. Claude, Anthropic’s AI, turns that into a draft. I read it, tell Claude what’s wrong, and Claude revises. We repeat the loop until the draft says what I mean. Claude wrote most of the sentences. The ideas, the judgments, and the final approval are mine.

Sentences: mostly Claude's

Draft-first

I write the draft myself, then Claude and I work through several recursive rounds. Claude gives substantial feedback, I revise, or ask Claude to, and Claude reads it again, until I’m satisfied. The sentences are mostly mine. Claude’s role is reader and editor.

Sentences: mostly mine

Co-writing

Claude and I trade passages. I write a section, Claude drafts the next, and each of us revises the other’s. The sentences are a blend, and so are the decisions.

Sentences: a blend

Solo

I write and revise it myself, and none of my own words on the page come from AI. Using AI for simple research, a spelling and grammar check, or brief feedback doesn’t count. That isn’t generative work, and none of it ends up in the piece. The one exception is brief research, which I state clearly and cite following APA or MLA guidelines.

Sentences: all mine

None of these is the “right” way. They’re different tools for different pieces, and the label tells you which one I used. In every mode, I’m responsible for what I publish.

Why labels instead of a citation

Academic writing already has a way to credit AI: cite it like a source. APA and MLA both have formats for it. You name the tool, sometimes the prompt, and the date.

That’s a start, but it leaves a lot to be desired. A citation treats AI like something you consulted once, either first, to generate a draft, or last, to polish one. It doesn’t show the middle, where most of the writing happens. It can’t tell a reader who drafted, who judged, or how many times the piece went back and forth.

That middle is what I think we should be reflecting on, as writers and as teachers. So instead of a citation, each essay here gets a label for the kind of partnership it was, and a note at the end on how it actually went. The labels and the notes are my answer to what the citation leaves out.

Why not journals, for now

This fall I’m stepping back from academic and literary publishing. Part of it is time: I’m teaching a full load, and the submission cycle is a grind. Part of it is that some journals, especially in creative writing, now decline any work made with AI, even when it’s disclosed.

I understand the concern, and I’m not going to fight it right now. As a professor, I have a responsibility to teach my college students how to use AI ethically and effectively, and I can’t teach what I haven’t practiced. So I’ve had to learn it myself, in my own writing: recursively, ethically, and with every decision still mine. Figuring that out, for myself and for my students, is the work I need to be doing, not arguing with journals about their fears of the future.

So I’m publishing here first, openly.

What that looks like

Each essay is labeled with how it was made.
Works in progress are marked, and I revise them in public.
I’m responsible for every sentence I publish, whoever typed it.